Message-Id: <199510040612.CAA09433@CS.UTK.EDU>
Date:        Wed, 4 Oct 1995 06:34:44 +0100
From: Eric Thomas <ERIC@SEARN.SUNET.SE>
Subject:     Re: What's the sender header for?
To: drums@CS.UTK.EDU, Eric Norman <ejnorman@macc.wisc.edu>
In-Reply-To: Message of Tue, 3 Oct 95 23:08:07 -0500 from Eric Norman
             <ejnorman@macc.wisc.edu>

On  Tue, 3  Oct 95  23:08:07 -0500  Eric Norman  <ejnorman@macc.wisc.edu>
said:

>Yes, there may  indeed me more than one AGENT  involved; the whole point
>was really  that the Sender:  header should reflect whomever  caused the
>message to be sent (to those recipients) and that the sending AGENT is
>*not* the group identified as drums@cs.utk.edu.

My  point  was that  the  sending  AGENT  for  LISTSERV lists  is  always
LISTSERV@you.name.it. It's the same for all  the lists on any given host,
and this is a technical fact.  Unfortunately, it's a rather useless thing
to put in a "Sender:" field since the whole purpose of the exercise is to
let people know which list it  comes from, for their personal convenience
in sorting and managing their mail. So LISTSERV uses the list name. Quite
frankly, 9 years  after the fact I don't  care if a case can  be built to
argue that  it should read LISTSERV@PEACH.EASE.LSOFT.COM  instead. It's a
useless value to put in that  field and people don't install updates that
turn something useful into something  useless, so I don't bother thinking
about them.

>So what are you  suggesting here? Should it be possible  to tell from an
>address whether  it represents  a carbon-based  or a  silicon-based life
>form?

Yep. The problem  is that the silicon-based life forms  are too stupid to
be  entrusted with  anything requiring  the use  of a  brain. Very  often
people write  to the  carbon life  forms to explain  that they've  had no
success getting the silicon life forms to understand that they want to do
this or that. The last thing they expect is yet another silicon screwup.

"Hello and welcome  to Barfproc 2.001. After reading  your message titled
'Can SOMEBODY please HELP???', I have determined that it was a misspelled
GETLOST  Barfproc command.  For your  convenience, I  have replaced  your
entire message  with a  GETLOST request and  discarded the  original. The
results  from your  GETLOST request  will follow  in a  separate message.
Please do not re-issue  a GETLOST request; I have taken  care of this for
you. You do not need to do anything. Barfproc's MindReader(TM) technology
has taken care of everything for you. Just sit back and relax, everything
is under control. Thank you for your cooperation".

>Are you suggesting that automatic  deletion via mail messages should not
>be allowed?

I'm suggesting that if people want to send a silicon command, they should
write  to  a silicon  address,  not  write to  a  carbon  address on  the
assumption that there will be some piece of broken silicon trying to do a
passable job  at filtering the requests.  Of course there will  always be
people who  write to the  wrong address and list  owners are free  to use
their  MUA's filtering  capabilities  to screen  the  requests. I'm  just
saying that  the whole concept of  sending a computer command  to a human
person on  the assumption that  he'll have gotten  so fed up  with having
computer commands sent to him that he'll have developed a computer filter
to sort the junk  from the real human mail is broken.  And it's even more
broken to  make it work,  because that's the best  way to teach  users to
continue doing that. This is an  old controversial debate that keep being
reopened from  time to time and  doesn't seem to get  anywhere. There's a
camp of  people who like  to think they're  capable of writing  code that
will tell the real, "meaningful" human mail from the "Please Sir, I would
like to stop getting the mail from  your list because my mailbox is full,
thank you!" requests,  and that the reason people like  me oppose this is
that we're too  stupid to write such powerful and  advanced software. And
then  there's  people like  me  who  like to  think  that  the time  when
computers  are able  to do  that kind  of work  reliably in  50+ Internet
languages has yet to come and  that meanwhile computers are stressful and
frustrating  enough  for  non-technical  users that  there  should  be  a
standard address to which they can write, safe in the knowledge that they
will be talking to a human being.

I don't  know, would you  ever consider running software  on "postmaster"
that says "Hey, it sounds like you  were asking for the e-mail address of
CASINI, ? so here  it is ---> pcasini@xyz.edu <--- We  hope you are happy
with  this  superb improved  high-tech  AI  service from  the  postmaster
address!  If your  query  was about  something else,  just  resend it  to
real-postmaster@xyz.edu. What, you didn't save  a copy? HAHAHAHAHA! I bet
now you will!" Setting  aside the wording :-), this is  the kind of stuff
I've seen  from the  most custom-improved  xxx-request filters,  the ones
that are  the most "efficient" in  cutting out the "junk  requests". Hmm,
one of them actually *did* say "For the stupid, [the above] means..."

>I sure don't  see how you can  claim that sending bounces  to Sender: is
>breakage since it's specifically allowed by RFC822. It'll be breakage if
>we eliminate this, but it isn't now.

It's a breakage for SMTP software, and that's nothing new.

  Eric